A Simulated Annealing-Based Approach for the Optimization of Routine Maintenance Interventions

Metaheuristics are often adopted to solve optimization problems where some requests need to be scheduled among a finite number of resources, i.e., the so called scheduling problems. Such techniques approach the optimization problems by taking inspiration from a certain physical phenomenon. Simulated annealing is a metaheuristic approach inspired to the controlled cooling of a material from a high temperature to a state in which internal defects of the crystals are minimized. In this paper, we use a simulated annealing-based approach to solve the problem of the scheduling of geographically distributed routine maintenance interventions. Each intervention has to be assigned to a maintenance team and the choice among the available teams and the order in which interventions are performed by each team are based on team skills, cost of overtime work, and cost of transportation. We compare our solution algorithm versus an exhaustive approach. First, we consider a real industrial use case and show several numerical results to analyze the effect of the parameters of the simulated annealing on the accuracy of the solution and on the execution time of the algorithm. Then, we provide results varying the parameters and dimension of the considered problem highlighting how they affect reliability and efficiency of our algorithm.